ELDER MISTREATMENT: A ROAD TO A NEW RISK ALGORITHM USING ICD-10 DATA

Elder mistreatment (EM)-a growing public health problem in the United States-is associated with poor quality of life, accelerated disability, and premature death. EM remains underrecognized and underreported, with the diagnosis frequently being made at a late stage when the afflicted older adults have experienced prolonged suffering including emotional, physical, and financial abuse. One way to increase early EM recognition and timely management is through the development and implementation of predictive algorithms that can help alert clinicians to the need for additional screening. In earlier work, using International Classification of Diseases, Tenth Revision (ICD-10) diagnostics codes, we were able to develop a logistic regression approach to determine an EM predictive algorithm with an area under the curve (AUC) of 0.73, a sensitivity of 80%, and specificity of 47%. Subsequently, this model was applied to a vulnerable group of older adults-those with Alzheimer's disease and related dementias (ADRD). In this cohort, we found illiteracy to be the strongest predictor associated with EM. These EM predictive algorithms can be useful in the creation of electronic health record (EHR) reminders to prompt clinicians to suspect EM and implement appropriate follow-up care plans.